Accurate parameter identification is essential for reliable photovoltaic (PV) modeling. Although metaheuristic algorithms have been widely used for this task, many methods still suffer from slow convergence and suboptimal solutions. To improve both accuracy and efficiency, we propose a two-stage framework that combines maximum power matching (MPM) with a multistrategy enhanced northern goshawk optimization (MSENGO) algorithm. First, the measured current–voltage ( I – V ) data are preprocessed to remove outliers and reduce redundancy. Next, MPM provides initial parameter estimates, which are then refined by MSENGO. MSENGO incorporates three complementary mechanisms: tent-chaotic initialization for enhancing population diversity, a wavelet-based mutation operator for intensified local refinement, and a nonlinear time-varying coordination schedule (sine-decreasing and cosine-increasing) to adaptively regulate the exploration–exploitation trade-off. On the CEC2017 benchmark set (F1–F12), MSENGO attains the theoretical optimum on 11 out of 12 functions and exhibits faster convergence than the compared optimizers. For PV parameter identification under three irradiance levels (379, 590, and 900 W/m 2 ), MSENGO achieves root mean square error (RMSE) values of 0.00717, 0.00655, and 0.00651 A, respectively, with R 2 = 0.9999 in all cases and computation times of 12.15–13.02 s. Compared with the best baseline method in each irradiance case, the RMSE reduction reaches approximately 12–48%, demonstrating clear accuracy and efficiency advantages. The proposed framework also maintains competitive performance when extended to more complex PV models (double-diode model and triple-diode model), indicating good generality.
Liu et al. (Mon,) studied this question.